Ward Cunningham coined the technical debt metaphor to describe the cost of expedient engineering decisions that create future rework: code written quickly to meet a deadline that works but is hard to change, dependencies that were convenient at the time but create coupling, a schema designed for the initial use case that does not accommodate the evolved requirements. Like financial debt, technical debt is not inherently wrong — it can be the right trade-off when speed creates competitive advantage. The problem is unacknowledged, unmanaged technical debt that compounds silently until it dominates development velocity.
There are three distinct types of technical debt, and conflating them leads to poor management decisions. Deliberate debt is consciously chosen — the team knows a better approach exists but chooses the expedient one for valid business reasons. Accidental debt is created through ignorance — the team was not aware of a better practice when making the decision. Bit rot is debt that accumulates passively as the environment around the code changes: a library with known vulnerabilities, a deprecated API, an approach that was valid three years ago but is now an anti-pattern.
Managing technical debt starts with making it visible. A debt registry — a tracked list of significant technical debt items with an estimate of the cost they are imposing on development velocity — transforms an invisible problem into a managed one. Teams that maintain a debt registry can have honest conversations with product management about the trade-off between investing in debt reduction versus investing in new features.
The most effective debt reduction practice is the "boy scout rule": always leave the code cleaner than you found it. When working in a section of code to implement a feature, improve the surrounding code as you go — improve naming, extract functions, add a test. This distributed, continuous improvement prevents debt from accumulating to the point of requiring a big-bang refactoring effort.
Reserve capacity for debt reduction: ten to twenty percent of engineering time dedicated to tech debt is sustainable and delivers compounding returns in velocity.